An Active Noise Control Algorithm Principle and Analysis without Secondary Path Identification Based on Kalman Filter

摘要

Most active noise control (ANC) algorithms require the identification of the secondary path, thus suffer from large complexity, increased residual noise power and algorithm divergence. In this paper, we propose a novel ANC algorithm without secondary path identification based on Kalman filter, referred to as Model Error Compensatory Kalman Filter (MECKF). The ANC problem is described in discrete-time state-space form first, then the dynamics of the primary path can be attributed to the state variables. Kalman filter is applied to estimate the weights using residual noise sequence. Furthermore, based on acoustics properties and stochastic theory, we introduce a model error compensating mechanism by shifting the influence of the unknown secondary path into variance of measurement matrix. In addition, a new method of estimating the statistical properties of the noise in dynamic model is given in the context of ANC system, with merits of reduced computational complexity, increased convergence rate, and ensured real-time.